From a266fe3eca52c00e0c32abebeccff027df6845c8 Mon Sep 17 00:00:00 2001 From: Askar Dyussekeyev <91150176+dyussekeyev@users.noreply.github.com> Date: Wed, 11 Nov 2026 00:34:32 +0500 Subject: [PATCH 1/5] Upgrade the connector base layer so the release scan can pass MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The Release run for v0.2.4 failed on the image scan for all six connectors. Their shared python:3.12-slim base carries a Debian snapshot whose zlib1g (1:1.3.dfsg+really1.3.1-1+b1) has two HIGH findings with a fix already published as +deb13u1, and the gate fails a fixable HIGH rather than shipping it. backend/Dockerfile already upgrades its base layer for exactly this reason; the connectors never did. The upgrade runs before the packages each image installs, so a pinned version such as dnstwist=0~20250130-1 is still what apt applies last and a rebuild cannot move it. Verified by building all six images locally (zlib1g becomes +deb13u1, dnstwist still reports 20250130) and by running Trivy 0.69.0 with the workflow's own flags on connector-hibp: exit 0. 🤖 Generated with Codebuff Co-Authored-By: Codebuff --- connectors/dns_security/Dockerfile | 4 ++++ connectors/dnstwist/Dockerfile | 5 +++++ connectors/exposure_endpoints/Dockerfile | 13 +++++++++++++ connectors/hibp/Dockerfile | 12 ++++++++++++ connectors/intelligence/Dockerfile | 13 +++++++++++++ connectors/shodan/Dockerfile | 12 ++++++++++++ 6 files changed, 59 insertions(+) diff --git a/connectors/dns_security/Dockerfile b/connectors/dns_security/Dockerfile index 710a870..d0a787f 100644 --- a/connectors/dns_security/Dockerfile +++ b/connectors/dns_security/Dockerfile @@ -11,7 +11,11 @@ RUN pip install --no-cache-dir --require-hashes -r /app/requirements.txt # against the system store is the point of the authoritative pass, and `%{certs}` # needs curl 7.88 (the worker records its version and refuses to probe on an older # build rather than reporting hosts as certificate-less). +# The base layer is upgraded before the probe's packages are installed; the +# reason is in `connectors/hibp/Dockerfile`. It keeps whether this image passes +# the release scan from depending on when Docker last rebuilt the base. RUN apt-get update \ + && apt-get upgrade -y --no-install-recommends \ && apt-get install -y --no-install-recommends curl ca-certificates \ && rm -rf /var/lib/apt/lists/* COPY connectors/dns_security/main.py connectors/dns_security/tls_probe.py /app/ diff --git a/connectors/dnstwist/Dockerfile b/connectors/dnstwist/Dockerfile index 51d9c76..1f6d075 100644 --- a/connectors/dnstwist/Dockerfile +++ b/connectors/dnstwist/Dockerfile @@ -14,7 +14,12 @@ FROM python:3.12-slim # for the same domain — a change that would otherwise arrive silently in a rebuild. # The pin is a Debian version; a base-image bump whose archive serves another one # fails the build rather than the scan. +# The base layer is upgraded before the pin is installed (reason: +# `connectors/hibp/Dockerfile`), so the pinned install is the last thing apt +# does here and the upgrade cannot move dnstwist off the release that +# `_EXPECTED_DNSTWIST_VERSION` names. RUN apt-get update \ + && apt-get upgrade -y --no-install-recommends \ && apt-get install -y --no-install-recommends dnstwist=0~20250130-1 python3-dnspython \ && rm -rf /var/lib/apt/lists/* diff --git a/connectors/exposure_endpoints/Dockerfile b/connectors/exposure_endpoints/Dockerfile index 1d36974..d8f9c71 100644 --- a/connectors/exposure_endpoints/Dockerfile +++ b/connectors/exposure_endpoints/Dockerfile @@ -1,4 +1,17 @@ FROM python:3.12-slim + +# The base layer is upgraded, not just the packages this image asks for. The +# official `python:3.12-slim` carries a Debian trixie snapshot, and its +# point-release security fixes reach the archive only after Docker rebuilds that +# image: at 0.2.4 the snapshot shipped a `zlib1g` that Trivy reports as HIGH +# (CVE-2026-76844, CVE-2026-85091) with a fix already published as `+deb13u1`. +# The release workflow fails on a fixable HIGH, so nothing ships until this +# layer moves. `upgrade`, not `dist-upgrade`: only packages the base image +# already installed are moved, and no new package is added. +RUN apt-get update \ + && apt-get upgrade -y --no-install-recommends \ + && rm -rf /var/lib/apt/lists/* + WORKDIR /app COPY connectors/base /opt/sdk RUN pip install --no-cache-dir /opt/sdk diff --git a/connectors/hibp/Dockerfile b/connectors/hibp/Dockerfile index 07935cd..0693e30 100644 --- a/connectors/hibp/Dockerfile +++ b/connectors/hibp/Dockerfile @@ -1,5 +1,17 @@ FROM python:3.12-slim +# The base layer is upgraded, not just the packages this image asks for. The +# official `python:3.12-slim` carries a Debian trixie snapshot, and its +# point-release security fixes reach the archive only after Docker rebuilds that +# image: at 0.2.4 the snapshot shipped a `zlib1g` that Trivy reports as HIGH +# (CVE-2026-76844, CVE-2026-85091) with a fix already published as `+deb13u1`. +# The release workflow fails on a fixable HIGH, so nothing ships until this +# layer moves. `upgrade`, not `dist-upgrade`: only packages the base image +# already installed are moved, and no new package is added. +RUN apt-get update \ + && apt-get upgrade -y --no-install-recommends \ + && rm -rf /var/lib/apt/lists/* + WORKDIR /app # Shared SDK first (cached layer). diff --git a/connectors/intelligence/Dockerfile b/connectors/intelligence/Dockerfile index 34a386d..dab1a8b 100644 --- a/connectors/intelligence/Dockerfile +++ b/connectors/intelligence/Dockerfile @@ -1,4 +1,17 @@ FROM python:3.12-slim + +# The base layer is upgraded, not just the packages this image asks for. The +# official `python:3.12-slim` carries a Debian trixie snapshot, and its +# point-release security fixes reach the archive only after Docker rebuilds that +# image: at 0.2.4 the snapshot shipped a `zlib1g` that Trivy reports as HIGH +# (CVE-2026-76844, CVE-2026-85091) with a fix already published as `+deb13u1`. +# The release workflow fails on a fixable HIGH, so nothing ships until this +# layer moves. `upgrade`, not `dist-upgrade`: only packages the base image +# already installed are moved, and no new package is added. +RUN apt-get update \ + && apt-get upgrade -y --no-install-recommends \ + && rm -rf /var/lib/apt/lists/* + WORKDIR /app COPY connectors/base /opt/sdk RUN pip install --no-cache-dir /opt/sdk diff --git a/connectors/shodan/Dockerfile b/connectors/shodan/Dockerfile index 1de0c4f..43e84f1 100644 --- a/connectors/shodan/Dockerfile +++ b/connectors/shodan/Dockerfile @@ -1,5 +1,17 @@ FROM python:3.12-slim +# The base layer is upgraded, not just the packages this image asks for. The +# official `python:3.12-slim` carries a Debian trixie snapshot, and its +# point-release security fixes reach the archive only after Docker rebuilds that +# image: at 0.2.4 the snapshot shipped a `zlib1g` that Trivy reports as HIGH +# (CVE-2026-76844, CVE-2026-85091) with a fix already published as `+deb13u1`. +# The release workflow fails on a fixable HIGH, so nothing ships until this +# layer moves. `upgrade`, not `dist-upgrade`: only packages the base image +# already installed are moved, and no new package is added. +RUN apt-get update \ + && apt-get upgrade -y --no-install-recommends \ + && rm -rf /var/lib/apt/lists/* + WORKDIR /app # Shared SDK first (cached layer). From f49403beb4d3a74f3638a698e4b180efe78609b6 Mon Sep 17 00:00:00 2001 From: Askar Dyussekeyev <91150176+dyussekeyev@users.noreply.github.com> Date: Wed, 11 Nov 2026 00:46:21 +0500 Subject: [PATCH 2/5] Pull the release's profiled connector images by name MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The 0.2.4 release run built and promoted all eight images and then failed in "Start the published images" before the stack was raised: four connectors were reported missing on the runner although the release had published them. They are behind Compose profiles (`shodan`, `hibp`, `netlas`, `netlas-endpoints`, `shodan-endpoints`), and `docker compose pull` only pulls services of active profiles. It fetched six of the eight images without saying so, and the digest guard, which checks all eight, failed on the rest. The guard was right; the pull was wrong. The job now names the matrix's images once in `RELEASE_IMAGES`, pulls each by name, and the guard reads that list instead of keeping a second copy that can drift. `docker compose pull` still runs afterwards for the datastores. Verified by running the guard script extracted from the workflow against a stubbed `docker`: it passes for all eight and still fails, naming the image, when one is absent. The list matches the build matrix, and the release gates (47 tests) pass. 🤖 Generated with Codebuff Co-Authored-By: Codebuff --- .github/workflows/release.yml | 37 ++++++++++++++++++++++------------- 1 file changed, 23 insertions(+), 14 deletions(-) diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml index 5c08eaf..7953e9a 100644 --- a/.github/workflows/release.yml +++ b/.github/workflows/release.yml @@ -277,6 +277,14 @@ jobs: # replaces reached healthy in about two minutes; the bound is for the step # that stops returning, not for a healthy run. timeout-minutes: 25 + env: + # The images the release matrix publishes, named once for the whole job: the + # pull below and the digest guard after it read this list, so a connector + # added to the matrix is one edit here rather than two that can drift. It is + # the matrix that is authoritative; this is the copy the job can iterate. + RELEASE_IMAGES: >- + backend frontend connector-dnstwist connector-shodan connector-hibp + connector-dns-security connector-intelligence connector-exposure-endpoints permissions: contents: read # The images are private: reading them back is the whole exercise, and the @@ -321,10 +329,20 @@ jobs: password: ${{ secrets.GITHUB_TOKEN }} - name: Pull every image the release published - # Every image the matrix names, including the connectors this job does not - # start: a tag that resolves to a missing manifest is a release an - # operator cannot install, and `pull` is the cheapest place to find out. - run: docker compose pull + # Pulled by name, not left to `docker compose pull`. Compose only pulls the + # services of active profiles, and four of the six connectors sit behind + # profiles (`shodan`, `hibp`, `netlas`, `netlas-endpoints`, + # `shodan-endpoints`): with no profile enabled it fetched six images of the + # eight the matrix publishes, silently, and the guard below then reported + # four published images as missing. That is how this job failed on its + # first real run, at 0.2.4. Compose still pulls afterwards for the + # datastores the stack starts with. + run: | + set -euo pipefail + for image in $RELEASE_IMAGES; do + docker pull "${OPENDRP_IMAGE_PREFIX}/${image}:${OPENDRP_VERSION}" + done + docker compose pull - name: Refuse an image that was built instead of pulled # The smoke has to test the artifact and not the source tree. A locally @@ -339,16 +357,7 @@ jobs: version = sys.argv[1] prefix = os.environ["OPENDRP_IMAGE_PREFIX"] - images = [ - "backend", - "frontend", - "connector-dnstwist", - "connector-shodan", - "connector-hibp", - "connector-dns-security", - "connector-intelligence", - "connector-exposure-endpoints", - ] + images = os.environ["RELEASE_IMAGES"].split() problems = [] for image in images: reference = f"{prefix}/{image}:{version}" From ffd98ad68156fb2f2efafb40fdd0d376463917ae Mon Sep 17 00:00:00 2001 From: Askar Dyussekeyev <91150176+dyussekeyev@users.noreply.github.com> Date: Wed, 11 Nov 2026 00:52:03 +0500 Subject: [PATCH 3/5] Gate the smoke job's image list against the build matrix MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The smoke job now names the images it pulls in `RELEASE_IMAGES`, and the digest guard reads that list. Nothing stopped the list from drifting from the matrix it copies: the 0.2.4 run failed because the pull and the guard each kept a list of their own, and only one of them matched what Compose had fetched. `validate_published_image_list` reads the matrix and the job's own env block and reports an image missing from the list, an image the matrix does not produce, a repeated image, a job with no list at all, and a list no step reads. It is wired into the existing `check_release_images.py` entry point, which CI already runs before the build and the tagged workflow runs in `verify-version`, so the gate needs no new invocation. The tests mutate the shipped workflow: dropping `connector-hibp` from the list, adding an unpublished image, repeating one, renaming the variable, hardcoding the pull, and removing the job each have to be reported. 🤖 Generated with Codebuff Co-Authored-By: Codebuff --- backend/tests/test_release_image_matrix.py | 63 +++++++++++++ scripts/check_release_images.py | 101 ++++++++++++++++++++- 2 files changed, 163 insertions(+), 1 deletion(-) diff --git a/backend/tests/test_release_image_matrix.py b/backend/tests/test_release_image_matrix.py index 2806407..09507c4 100644 --- a/backend/tests/test_release_image_matrix.py +++ b/backend/tests/test_release_image_matrix.py @@ -210,3 +210,66 @@ def test_dockerfile_outside_repository_is_reported_from_config() -> None: assert any("connector-intelligence" in failure and "context/dockerfile" in failure for failure in failures) +def test_smoke_job_pulls_every_image_the_matrix_builds() -> None: + assert _GATE.validate_published_image_list(_release()) == [] + + +def test_smoke_image_list_missing_a_matrix_image_is_reported() -> None: + source = _release().replace( + "connector-dnstwist connector-shodan connector-hibp", + "connector-dnstwist connector-shodan", + ) + + failures = _GATE.validate_published_image_list(source) + + assert any("connector-hibp" in failure and "missing" in failure for failure in failures) + + +def test_smoke_image_list_naming_an_unpublished_image_is_reported() -> None: + source = _release().replace( + "backend frontend connector-dnstwist", + "backend frontend connector-otherwise-unpublished", + ) + + failures = _GATE.validate_published_image_list(source) + + assert any("connector-otherwise-unpublished" in failure for failure in failures) + + +def test_smoke_image_list_repeating_an_image_is_reported() -> None: + source = _release().replace( + "backend frontend connector-dnstwist", + "backend frontend backend connector-dnstwist", + ) + + failures = _GATE.validate_published_image_list(source) + + assert any("repeats" in failure and "backend" in failure for failure in failures) + + +def test_smoke_job_without_the_shared_list_is_reported() -> None: + source = _release().replace(" RELEASE_IMAGES: >-\n", " SOMETHING_ELSE: >-\n") + + failures = _GATE.validate_published_image_list(source) + + assert any("RELEASE_IMAGES" in failure for failure in failures) + + +def test_smoke_job_whose_list_no_step_reads_is_reported() -> None: + source = _release().replace( + "for image in $RELEASE_IMAGES; do", "for image in backend frontend; do" + ).replace('images = os.environ["RELEASE_IMAGES"].split()', 'images = ["backend"]') + + failures = _GATE.validate_published_image_list(source) + + assert any("no step reads it" in failure for failure in failures) + + +def test_smoke_job_missing_entirely_is_reported() -> None: + source = _release().replace(" smoke-published-images:\n", " start-images:\n") + + failures = _GATE.validate_published_image_list(source) + + assert any("no smoke-published-images job" in failure for failure in failures) + + diff --git a/scripts/check_release_images.py b/scripts/check_release_images.py index 6dc410a..eab490f 100644 --- a/scripts/check_release_images.py +++ b/scripts/check_release_images.py @@ -84,6 +84,51 @@ def finish() -> None: return entries, failures +_FOLDED_SCALAR = {">", ">-", ">+", "|", "|-", "|+"} + + +def _job_block(source: str, job: str) -> list[str]: + """The lines of one job, from its `key:` line up to the next job.""" + block: list[str] = [] + in_job = False + for line in source.splitlines(): + stripped = line.strip() + indent = len(line) - len(line.lstrip()) + if indent == 2 and stripped.endswith(":") and not stripped.startswith("#"): + in_job = stripped[:-1] == job + continue + if in_job: + block.append(line) + return block + + +def _job_env_list(block: list[str], key: str) -> tuple[list[str], bool]: + """Read a whitespace-separated list from a job's `env:` block.""" + values: list[str] = [] + found = False + in_env = False + collecting = False + for line in block: + stripped = line.strip() + indent = len(line) - len(line.lstrip()) + if collecting: + if stripped and indent > 6 and not stripped.startswith("#"): + values.extend(stripped.split()) + continue + collecting = False + if indent == 4 and stripped and not stripped.startswith("#"): + in_env = stripped == "env:" + continue + if in_env and indent == 6 and stripped.startswith(f"{key}:"): + found = True + value = stripped.split(":", 1)[1].strip() + if value in _FOLDED_SCALAR: + collecting = True + else: + values = value.split() + return values, found + + def _repo_path(path: str, root: str | Path = ROOT) -> str: """Normalize POSIX or Windows paths to repo-relative POSIX form.""" root_path = str(root) @@ -214,6 +259,56 @@ def validate_release_coverage(release_source: str, compose_config: dict) -> list return failures +def validate_published_image_list(release_source: str) -> list[str]: + """Fail when the smoke job's image list drifts from the build matrix. + + The smoke job pulls the images it starts by name, and the digest guard after + it reads the same list. Both used to keep their own copy of the matrix's + images, which is how the job's first real run at 0.2.4 asked the runner for + four images nobody had pulled. The matrix stays the authority; this checks + that the list the job iterates is exactly the matrix, with nothing missing + and nothing invented. + """ + matrix, failures = _release_matrix(release_source) + block = _job_block(release_source, "smoke-published-images") + if not block: + failures.append( + "release workflow has no smoke-published-images job, so nothing pulls " + "the images the release publishes" + ) + return failures + listed, found = _job_env_list(block, "RELEASE_IMAGES") + if not found or not listed: + failures.append("the smoke job names no images: RELEASE_IMAGES is missing or empty") + return failures + duplicates = sorted({image for image in listed if listed.count(image) > 1}) + if duplicates: + failures.append("RELEASE_IMAGES repeats images: " + ", ".join(duplicates)) + unknown = sorted(set(listed) - set(matrix)) + if unknown: + failures.append( + "RELEASE_IMAGES names images the build matrix does not produce: " + + ", ".join(unknown) + ) + missing = sorted(set(matrix) - set(listed)) + if missing: + failures.append( + "RELEASE_IMAGES is missing images the build matrix publishes: " + + ", ".join(missing) + ) + uses = [ + line + for line in block + if "RELEASE_IMAGES" in line and not line.strip().startswith("RELEASE_IMAGES:") + ] + if not uses: + failures.append( + "RELEASE_IMAGES is declared but no step reads it, so the pull is not " + "the list this gate checked" + ) + return failures + + def main(argv: list[str] | None = None) -> int: parser = argparse.ArgumentParser(description=__doc__) parser.add_argument("compose_json", type=Path, help="resolved `docker compose config --format json` output") @@ -225,12 +320,16 @@ def main(argv: list[str] | None = None) -> int: print(f"release image gate: FAILED — {exc}") return 2 failures = validate_release_coverage(release, compose) + failures += validate_published_image_list(release) if failures: print("release image gate: FAILED") for failure in failures: print(f" {failure}") return 1 - print("release image gate: OK — every resolved Compose image has a matching release build") + print( + "release image gate: OK — every resolved Compose image has a matching " + "release build, and the smoke job pulls exactly that set" + ) return 0 From e93e610919e2ecff88e71246f62e4c164c402def Mon Sep 17 00:00:00 2001 From: Askar Dyussekeyev <91150176+dyussekeyev@users.noreply.github.com> Date: Wed, 11 Nov 2026 00:55:06 +0500 Subject: [PATCH 4/5] Take the smoke job's pull list from Compose MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The pull step named the eight images itself, which is a second copy of what `docker-compose.yml` already says and one more place for a new connector to be forgotten. `docker compose --profile '*' config --images` resolves the image of every service instead, profiled connectors included, and that set is piped to `docker pull`. The wildcard is the point: the four connectors that failed at 0.2.4 sit behind `shodan`, `hibp`, `netlas`, `netlas-endpoints` and `shodan-endpoints`, and without it Compose lists six of the ten images. `RELEASE_IMAGES` stays, because the guard after the pull reads it and the matrix gate keeps it equal to the build matrix; only the pull stops keeping its own copy. A test pins the wildcard so a revert to plain `docker compose pull` is caught in the suite rather than on a release. Checked locally at 0.2.4: the pipeline resolves the eight published images plus `postgres:17-alpine` and `redis:7-alpine`; the matrix gate reports no failures; 23 tests in the file pass. 🤖 Generated with Codebuff Co-Authored-By: Codebuff --- .github/workflows/release.yml | 24 +++++++++++----------- backend/tests/test_release_image_matrix.py | 13 ++++++++++-- 2 files changed, 23 insertions(+), 14 deletions(-) diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml index 7953e9a..e6ee6cb 100644 --- a/.github/workflows/release.yml +++ b/.github/workflows/release.yml @@ -329,20 +329,20 @@ jobs: password: ${{ secrets.GITHUB_TOKEN }} - name: Pull every image the release published - # Pulled by name, not left to `docker compose pull`. Compose only pulls the - # services of active profiles, and four of the six connectors sit behind - # profiles (`shodan`, `hibp`, `netlas`, `netlas-endpoints`, - # `shodan-endpoints`): with no profile enabled it fetched six images of the - # eight the matrix publishes, silently, and the guard below then reported - # four published images as missing. That is how this job failed on its - # first real run, at 0.2.4. Compose still pulls afterwards for the - # datastores the stack starts with. + # Compose is the list: `config --images` resolves the image of every + # service, and `--profile '*'` turns on the profiles the connectors sit + # behind (`shodan`, `hibp`, `netlas`, `netlas-endpoints`, + # `shodan-endpoints`). Without that wildcard `docker compose pull` fetched + # only the six images of the active profiles and said nothing, and the + # guard below reported four published images as missing — how this job + # failed on its first real run, at 0.2.4. Reading the set from Compose also + # means a new connector is covered by the Compose file alone, rather than + # by a second list in this workflow that has to be kept in step. run: | set -euo pipefail - for image in $RELEASE_IMAGES; do - docker pull "${OPENDRP_IMAGE_PREFIX}/${image}:${OPENDRP_VERSION}" - done - docker compose pull + docker compose --profile '*' config --images \ + | sort -u \ + | xargs -r -n1 docker pull - name: Refuse an image that was built instead of pulled # The smoke has to test the artifact and not the source tree. A locally diff --git a/backend/tests/test_release_image_matrix.py b/backend/tests/test_release_image_matrix.py index 09507c4..3025748 100644 --- a/backend/tests/test_release_image_matrix.py +++ b/backend/tests/test_release_image_matrix.py @@ -257,14 +257,23 @@ def test_smoke_job_without_the_shared_list_is_reported() -> None: def test_smoke_job_whose_list_no_step_reads_is_reported() -> None: source = _release().replace( - "for image in $RELEASE_IMAGES; do", "for image in backend frontend; do" - ).replace('images = os.environ["RELEASE_IMAGES"].split()', 'images = ["backend"]') + 'images = os.environ["RELEASE_IMAGES"].split()', 'images = ["backend"]' + ) failures = _GATE.validate_published_image_list(source) assert any("no step reads it" in failure for failure in failures) +def test_smoke_pull_asks_compose_for_every_profile() -> None: + """Plain `docker compose pull` skips the profiled connectors, which is the + failure this step exists to prevent.""" + smoke = "\n".join(_GATE._job_block(_release(), "smoke-published-images")) + + assert "--profile '*'" in smoke + assert "config --images" in smoke + + def test_smoke_job_missing_entirely_is_reported() -> None: source = _release().replace(" smoke-published-images:\n", " start-images:\n") From c47ddb2c65481f1ecc59d8e0c9383f92cd119b1c Mon Sep 17 00:00:00 2001 From: Askar Dyussekeyev <91150176+dyussekeyev@users.noreply.github.com> Date: Wed, 11 Nov 2026 00:56:44 +0500 Subject: [PATCH 5/5] Record the base layer and smoke-job fixes in the 0.2.4 notes MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit The release notes are published from this file, and v0.2.4 has not been published yet: both fixes ship in it, so they belong in its section rather than in Unreleased. One entry covers the connector base layer the image scan demanded, the other the pull that skipped the profiled connectors and the gate that now keeps its list equal to the build matrix. 🤖 Generated with Codebuff Co-Authored-By: Codebuff --- CHANGELOG.md | 24 ++++++++++++++++++++++++ 1 file changed, 24 insertions(+) diff --git a/CHANGELOG.md b/CHANGELOG.md index 83b9201..960ad2b 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -23,6 +23,30 @@ with it. `/api/v1/ready` to be ready and the frontend to serve the SPA mount point. The release notes are published only after it passes. +### Fixed + +- The connector images upgrade their Debian base layer before installing their + own packages. All six build `FROM python:3.12-slim`, and the snapshot that + image carried at 0.2.4 shipped a `zlib1g` (`1:1.3.dfsg+really1.3.1-1+b1`) with + two HIGH findings whose fix was already published as `+deb13u1`. The release + workflow fails an image that carries a fixable HIGH, so the first tagged run of + 0.2.4 stopped at the image scan of every connector while backend and frontend + passed — the backend image had upgraded its base layer for exactly this reason + and the connectors never did. The upgrade runs before each image installs its + own packages, so a pinned package such as `dnstwist=0~20250130-1` is still what + apt applies last and a rebuild cannot move it. +- The release smoke job pulls every image the release publishes, not the six + Compose fetches by default. Four of the six connectors sit behind profiles + (`shodan`, `hibp`, `netlas`, `netlas-endpoints`, `shodan-endpoints`), and + `docker compose pull` skips an inactive profile without saying so: the job + fetched six of the eight images, and the digest guard after it — which checks + all eight — then reported four images the release had published as missing. + The pull now takes its list from `docker compose --profile '*' config + --images`, so a new connector is covered by the Compose file alone rather than + by a second list in the workflow, and `scripts/check_release_images.py` fails + CI if the list the guard reads stops matching the build matrix, so the two + copies that failed here cannot drift apart again. + ## [0.2.3] - 2026-11-05 ### Security